Individual differences in the subjective effects of drugs of abuse are partially under genetic control. These differences are believed to influence the risk for drug abuse and dependence. Mice can be used to model sensitivity to the effects of drugs and offer a number of advantages relative to studies in human populations. We are proposing to fine map quantitative trait loci (QTLs) affecting methamphetamine (MA) sensitivity using a 34th generation SM/J x LG/J advanced intercross line (AIL). An AIL is an outbred population that is derived from a cross between two inbred strains and maintained beyond the F2 generation. AILs are powerful populations for fine mapping of QTLs because each individual accumulates a large number of recombinations over the course of many generations. These recombinations break down linkage disequilibrium (LD) between the QTL and surrounding markers, allowing high resolution mapping of the variants that influence quantitative traits. High resolution mapping is critically important to the future of QTL studies. Presently less than 1% of all QTLs identified in mice have led to successful gene identification. This is because standard populations like F2 crosses and small RI panels are good for identifying broad regions that contain QTLs (coarse mapping), but have too much LD to permit fine mapping. Three main factors have precluded the wide-spread use of AIL for fine mapping. First, AIL take several years to create; we have overcome this difficulty by obtaining an AIL that is already at the 34th generation. Second, a large number of markers must be genotyped due to the large number of recombinations; modern high-throughput genotyping makes it possible to type thousands of markers rapidly and at relatively low cost. Third, the methods for the analysis of an AIL have yet to be developed; this proposal addresses this third and final obstacle. AILs have a complex pedigree structure that must be taken into account when analyzing the relationship between genotype and phenotype. No currently available methods properly account for the pedigree structure in these populations, which causes both false negative and false positive errors. We will adapt three variance component methods that are widely used in human genetics. The first method (association) will use a standard mixed model to regress phenotype on QTL genotype while accounting for the correlations due to relatedness among pedigree members. The second method (within-family association) will use the quantitative transmission disequilibrium test (QTDT) to condition on parental genotype, thus eliminating the need to explicitly model the pedigree structure. The third method (linkage) will use classical quantitative trait linkage analysis to model phenotypic similarity using the number of alleles shared identical by descent while explicitly modeling the pedigree structure. We will then use simulations of possible genotypes given the known pedigree structure to evaluate the statistical significance of our results. SM/J and LG/J show extremely divergent locomotor responses to 2 mg/kg MA (>5-fold difference; p<10-13), suggesting the presence of one or more major QTLs for this phenotype. [unreadable] [unreadable] [unreadable]